If you use the Docker image, we highly recommend that you perform what amounts to a cluster restart upgrade. This process requires downtime, but takes very few steps and avoids problems with individual nodes rejoining the cluster and executing commands within containers.
The most important step is to leave your data volumes intact. Don’t run
docker-compose down -v.
Update the version strings in
docker-compose.yml. You can perform this step while the cluster is running. For more information, see Sample Docker Compose file.
Recreate the cluster using the updated file:
Wait for the cluster to start, and verify that your cluster returns the new version:
curl -XGET https://localhost:9200 -u 'admin:admin' -k
Verify cluster health and the expected number of containers and nodes:
curl -XGET https://localhost:9200/_cat/health?v -u 'admin:admin' -k docker ps
Open Kibana, and verify that your data is present.